This file contains mixture of analysis and forecast, retrieved from era interim dataset.
Analysis are available every six hours.
Forecast is run every 12 hours (00 and 12), forecast step is 3 hours.

If you take a look at the last column (fctstep), you will notice a bit "strange"
patern: 00 03 00 09... instead of 00 03 06 09. This is because forecast from 00h (fctstep=6) is replaced
with analysis. 
ECMWF operationaly runs model 2 times each day (00 and 12 h). Besides, they have analysis at 06h and 18h.
So we have replaced forecasted value at 06 AM (that comes from forecast start at 00) with analysis from 06am.
Next value(at 09h) is again forecast from 00AM run(fcstep=9). The same princip is applied for values at 18h and 21h,
but now the forecast run is at 12. It depends on you which data to use.

year  mm  dd  hh         tcc   fctst  fctstep
2013  05  10  00    0.182465     00     00     <==  forecast start 00, forecast step 00 - analysis 
2013  05  10  03    0.027222     00     03     <==  forecast start 00, forecast step 03 - forecast 
2013  05  10  06    0.007813     06     00     <==  analysis at    06  
2013  05  10  09    0.000061     00     09     <==  forecast start 00, forecast step 09 - forecast 
2013  05  10  12    0.012573     12     00     <==  forecast start 12, forecast step 00 - analysis 
2013  05  10  15    0.000336     12     03     <==  forecast start 12, forecast step 03 - forecast 
2013  05  10  18    0.012054     18     00     <==  analysis at    18

tcc (=) non-dimensional unit


----------header----------

year mm dd hh: VALID DATE
tcc: total cloud cover (ecmwf grib indicatorOfParameter 164.128)
fctst: forecast start hour
fctstep: forecast step